Radiocarbon constraints reveal time scales of soil carbon persistence

Abstract

Soils are currently a sink for atmospheric C, but may become a source in the coming decades. Predicting future gains or losses in soil C will require quantifying the time scales on which C cycles through soils, as well as deepening our understanding of the mechanisms controlling these cycling rates. Global patterns of soil C stocks and the radiocarbon (14C) signature of bulk soil C (∆14Cbulk) establish temperature as a master control on soil C ages and accumulation rates. Yet emerging understanding underscores the importance of mineral control for both soil C cycling rates and the temperature sensitivity of decomposition. The central aim of this dissertation is to quantify the time scales of soil C cycling on which mineralogical controls are relevant and the influence of the soil mineral assemblage on the responses of soil C ages and transit times to climate. Radiocarbon is a sensitive tracer for quantifying time scales of soil C cycling. The mean age of soil C can be constrained with observations of ∆14Cbulk, but the 14C signature of heterotrophically respired CO2 (∆14Crespired) adds a powerful constraint on the age of C returning to the atmosphere i.e., soil C transit time. Incubating archived soils would enable the construction of time series of ∆14Crespired, substantially reducing uncertainty from observations at single point in time. The objective of the first study in this dissertation (Ch. 2) is to assess the feasibility of measuring ∆14Crespired in archived soils by quantifying potential biases caused by air-drying, rewetting, and storage of soils prior to incubation. Results indicate storage has a negligible impact, but air-drying and rewetting leads to a small increase in the relative contribution of older C to respiration. However, the absolute bias in ∆14Crespired from air-drying and rewetting was minimal (±12‰ to ±40‰), suggesting that constructing time series of ∆14Crespired from incubations of archived soils is promising as long as soils undergo the same air-drying and rewetting procedure. In Ch. 3 of this dissertation, I compare the distribution and change over time in ∆14Cbulk and ∆14Crespired among soils developed on different parent materials (andesite, basalt, granite) but with similar mean annual soil temperature (MAST) and climate regime (warm ~ 12.0 °C, cool ~ 8.6 °C, cold ~ 6.6 °C) using archived soils. The results provide new evidence that mineral assemblages: 1) mediate climatic control of soil C turnover, and 2) are relevant for C cycling on annual to decadal time scales as well as centennial and longer. Furthermore, the effect of MAST on the change observed in ∆14Crespired over time was only significant in the soils with the lowest content of poorly crystalline metal (oxy) hydroxide (PCM) content, implying that soil organic matter interactions with these minerals may attenuate temperature sensitivity of soil C ages and transit times. Determining ages and transit times of soil C requires the use of a model. In Ch. 4 of this dissertation (Ch. 4) I demonstrate how time series of ∆14Crespired and 14Cbulk can be used to constrain soil C models using the data from Ch. 3. Different two-pool model structures yielded similar estimates for soil C ages, transit times, and inputs, indicating that 14Crespired and 14Cbulk are robust constraints for such a system. Trends in mean ages and transit times with respect to climatic and mineralogical factors were similar to those in ∆14Cbulk and ∆14Crespired, respectively. However, the models also yield probability distributions of age and transit time. The distributions reveal that in some soils, such as those with abundant PCMs, small amounts of highly ∆14C-depleted C can bias estimates of the mean, potentially leading to overestimates of ages or transit times. Modeled estimates of the pre-aging of soil C inputs show an increase with depth, adding to the growing recognition that observed increases in 14C age with depth may not be due solely to slower turnover, but also vertical transport. The central theme of this dissertation is that mineral-associated soil organic matter is not a homogenous pool, and in soils consisting of a wide range of soil mineral assemblages, consists of C cycling on time scales ranging from annual to millennial. Furthermore, ages and transit times of C in the PCM-rich soils of this study were less sensitive to temperature than in PCM-poor soils, highlighting the importance of accounting for mineral assemblages in predicting the effect of rising temperatures on soil C stocks.